Topics in Bioinformatics - Montefiore Institute ULgkbessonov/present_data/GBIO0009-1... · Topics...
Transcript of Topics in Bioinformatics - Montefiore Institute ULgkbessonov/present_data/GBIO0009-1... · Topics...
GBIO0009 - LECTURE 1
Topics in Bioinformatics
Kristel Van Steen, PhD2
Montefiore Institute - Systems and Modeling
GIGA - Bioinformatics
ULg
GBIO0009 - LECTURE 1
Lecture 1: Setting the pace
1 Evolving trends in bioinformatics
Definition
Historical notes
Challenges
2 Careers in bioinformatics
Topics in bioinformatics from a journal’s perspective
Becoming a bioinformatician
3 Gen-omics
Corner stone of modern bioinformatics
GBIO0009 - LECTURE 1
1 Evolving trends in bioinformatics
Definition
Bioinformatics as a field has arisen in parallel with the development of automated high-throughput methods of biological and biochemical discovery that yield a variety of forms of experimental data:
- DNA sequences, - gene expression patterns, - three-dimensional models of macromolecular structure, …
The field's rapid growth is driven by the vast potential for new understanding that can lead to new treatments, new drugs, new crops, and the general expansion of knowledge.
(http://findarticles.com/p/articles/mi_qa3886/is_200301/ai_n9182276/)
GBIO0009 - LECTURE 1
Oxford English Dictionary
(Molecular) bio– informatics: bioinformatics is conceptualising biology in terms of molecules (in the sense of physical chemistry) and applying "informatics techniques" (derived from disciplines such as applied maths, computer science and statistics) to understand and organize the information associated with these molecules, on a large scale. In short, bioinformatics is a management information system for molecular biology and has many practical applications. .
(see also: Luscombe et al. 2001 – review paper)
GBIO0009 - LECTURE 1
Towards a formal definition
Bioinformatics encompasses everything - from data storage - over data retrieval - to computational testing of biological hypotheses.
Bioinformatics information systems include - multi-layered software, - hardware, and - experimental solutions
bringing together a variety of tools and methods to analyze enormous quantities of “noisy” data.
(http://findarticles.com/p/articles/mi_qa3886/is_200301/ai_n9182276/)
GBIO0009 - LECTURE 1
Formal definition
Bioinformatics Computational biology Research, development or application of computational tools and approaches for expanding the use of biological, medical, behavioral or health data, including those to acquire, store, organize, analyze, or visualize such data
Development and application of data-analytical, theoretical methods, mathematical modeling and computational simulation to the study of biological, behavioral, and social systems.
(BISTIC Definition Committee, NIH, 2000)
GBIO0009 - LECTURE 1
Example: The TELEVIE PDAC-xome project
Setting / motivation
Pancreatic ductal adenocarcinoma (PDAC), the most common type of pancreatic
cancer, could be considered an orphan disease because, until present, it has not
been adopted by the pharmaceutical industry. Despite its relatively low population
incidence, it is the deadliest cancer worldwide with <6% 5-year survival rate. It is
highly plausible that complex interactions between multiple genomic, epigenomic
and environmental risk factors are involved
GBIO0009 - LECTURE 1
The PDAC-xome project
Aims and goals
Objective 1: to perform whole exome sequencing (WES) on selected samples from
the University Hospital in Liège (CHU-Liège) PDAC pathology repository and to
describe the mutational landscape of these patients.
Objective 2: to provide a molecular characterization of PDAC using the
aforementioned samples, hereby expanding the landscape of single nucleotide
polymorphisms in human PDAC to include rare variants while charting patient
heterogeneity.
Objective 3: to perform association analyses to identify candidate variants, genes,
and pathways involved in PDAC, possibly beyond main effects.
GBIO0009 - LECTURE 1
The PDAC-xome project - bioinformatics
Technology:
- platforms to
generate the data
- internal quality
control measures
- pipeline development
Genomics
Computation
Analysis
GBIO0009 - LECTURE 1
Computation - Analysis
The previous shows part of a general framework of “WES data analysis” (WES = whole exome sequencing) :
- raw reads QC, - preprocessing, - alignment, - post-processing, and - variant analysis (variant calling, annotation, and prioritization).
FASTQ, BAM, variant call format (VCF), and TAB (tab-delimited) refer to the standard file format of raw data, alignment, variant calls, and annotated variants, respectively. A selection of tools supporting each analysis step is shown in italic.
GBIO0009 - LECTURE 1
The PDAC-xome project – statistical genetics
- Data collection and WES analysis
- Molecular reclassification
- Association analysis
Collaborations between statistical geneticists, biomedics, medical doctors and
bioinformaticians (and many more) for this project, on the road to genome
and network medicine …
(Barabási et al. 2011)
GBIO0009 - LECTURE 1
Bioinformatics: opportunities in drug research development
Biofinformatics plays an increasing role in health care, for instance in new drug research and development:
- Identification of novel drug/vaccine targets - Structural predictions - Tapping into biodiversity - Reconstruction of metabolic pathways - Systems biology
GBIO0009 - LECTURE 1
Bioinformatics in the new millennium
Huge increase in available biological information
Classic paradigm of “molecular biology” is altering rapidly to “genomics”
Understanding of the new paradigms concerns more than simply “bench biology”
Discovery requires large scale systems and broad collaborations, and dealing with global problems as well
Funding comes in large amounts at group level, no longer a single laboratory or institution effort.
Accountable output
GBIO0009 - LECTURE 1
The pre-70’s: pioneering computational studies
Increased biological, structural understanding:
- Structure of DNA (Watson and Crick, 1953) - Encoding of genetic info for proteins (Gamow et al. 1956) - Structural properties of protein molecules (Anfinsen and Scheraga
1975) - Evolution of biochemical pathways (Horowitz 1945) - Gene regulation (Britten and Davidson 1969) - Chemical basis for development (Turing 1952)
Developments in fundamental computer science
- Information theory (Shannon and Weaver 1962) - Game theory (Neumann and Morgenstern 1953)
Combined: birth of computational biology
GBIO0009 - LECTURE 1
Bioinformatics Computational biology Research, development or application of computational tools and approaches for expanding the use of biological, medical, behavioral or health data, including those to acquire, store, organize, analyze, or visualize such data
Development and application of data-analytical, theoretical methods, mathematical modeling and computational simulation to the study of biological, behavioral, and social systems.
(BISTIC Definition Committee, NIH, 2000)
GBIO0009 - LECTURE 1
The pre-70’s: theoretical foundations
Merging of classical population genetics with molecular evolution (Ohta and Kimura 1971)
String comparison problem in computer science (Levin 1973)
The 80’s: More algorithms and resources
Development of efficient algorithms to cope with an increasing volume of information
Making available computer implementations for the wider scientific community
Overall: Shaping computational biology as an independent discipline
GBIO0009 - LECTURE 1
Completion of the Human Genome
Project
On June 26, 2000, the International Human
Genome Sequencing Consortium announced
the production of a
rough draft of the
human genome
sequence. An essentially
finished version is
announced by them in
April 2003.
GBIO0009 - LECTURE 1
2005+ : genomics and beyond
This review provides an introduction to the methods and problems tackled by bioinformatics.
In addition to identifying all the genes in genomes it is crucial to store and distribute the
information in databases. Annotation and identification of genes from genomes are crucial for
generating useful genome databases. Ethical, legal, and social implications of genome and gene
data are briefly discussed.
GBIO0009 - LECTURE 1
Challenges in bioinformatics
Data deluge (availability, what to archive and what not?, …)
Knowledge management (accessibility, usability, …)
Predicting, not just explaining (what comes first: hypothesis generation,
data collection? …)
Precision medicine (alias: personalized medicine; holistic approach –
correlating different causal associations – versus a reductionist approach –
targeting very specific biomarkers, negative gold standards … negative
controls)
Speciation (loss in biodiversity, evolutionary units, “integrative taxonomy”:
molecular, morphological, ecological and environmental information)
Inferring the tree of life (unresolved orthology assignment, gene sampling
pyramid
GBIO0009 - LECTURE 1
2 Careers in bioinformatics
Topics in bioinformatics from a journal’s perspective
(source: Scope of the journal “Bioinformatics”)
Data and Text Mining This category includes: New methods and tools for extracting biological information from text, databases and other sources of information. Description of tools to organize, distribute and represent this information. New methods for inferring and predicting biological features based on the extracted
information. The submission of databases and repositories of annotated text, computational tools and general methodology for the work in this area are encouraged, provided that they have been previously tested.
GBIO0009 - LECTURE 1
The journal ….
BioData Mining is an open access, peer reviewed, online journal encompassing research on all aspects of data mining applied to high-dimensional biological and biomedical data, focusing on computational aspects of knowledge discovery from large-scale genetic, transcriptomic, genomic, proteomic, and metabolomic data.
Topical areas include, but are not limited to:
Development, evaluation, and application of novel data mining and machine learning algorithms.
Adaptation, evaluation, and application of traditional data mining and machine learning algorithms.
Open-source software for the application of data mining and machine learning algorithms.
Design, development and integration of databases, software and web services for the storage, management, retrieval, and analysis of data from large scale studies.
Pre-processing, post-processing, modeling, and interpretation of data mining and machine learning results for biological interpretation and knowledge discovery.
GBIO0009 - LECTURE 1
Databases and Ontologies
This category includes: Curated biological databases, data warehouses, eScience, web services, database integration, biologically-relevant ontologies.
(Kim et al. 2012)
GBIO0009 - LECTURE 1
Bioimage Informatics This category includes novel methods for the acquisition, analysis and modeling of images produced by modern microscopy, with an emphasis on the application of innovative computational methods to solve challenging and significant biological problems at the molecular, sub-cellular, cellular, and tissue levels. This category also encourages large-scale image informatics methods/applications/software, joint analysis of multiple heterogeneous datasets that include images as a component, and development of bioimage-related ontologies and image retrieval methods.
GBIO0009 - LECTURE 1
Genome analysis This category includes: Comparative genomics, genome assembly, genome and chromosome annotation, identification of genomic features such as genes, splice sites and promoters.
GBIO0009 - LECTURE 1
Genetics and Population Analysis This category includes: Segregation analysis, linkage analysis, association analysis, map construction, population simulation, haplotyping, linkage disequilibrium, pedigree drawing, marker discovery, power calculation, genotype calling.
(Ziegler 2009) (Van Steen 2011)
GBIO0009 - LECTURE 1
Sequence analysis This category includes: Multiple sequence alignment, sequence searches and clustering; prediction of function and localisation; novel domains and motifs; prediction of protein, RNA and DNA functional sites and other sequence features.
GBIO0009 - LECTURE 1
Phylogenetics
This category includes: novel phylogeny estimation procedures for molecular data including nucleotide sequence data, amino acid data, whole genomes, SNPs, etc., simultaneous multiple sequence alignment and phylogeny estimation, phylogenetic approaches for any aspect of molecular sequence analysis (see Sequence Analysis scope), models of molecular evolution, assessments of statistical support of resulting phylogenetic estimates, comparative methods, coalescent theory, approaches for comparing phylogenetic trees, methods for testing and/or mapping character change along a phylogeny.
(Darwin 1859)
GBIO0009 - LECTURE 1
Gene Expression This category includes a wide range of applications relevant to the high-throughput analysis of expression of biological quantities, including microarrays (nucleic acid, protein, array CGH, genome tiling, and other arrays), RNA-seq, proteomics and mass spectrometry. Approaches to data analysis to be considered include statistical analysis of differential gene expression; expression-based classifiers; methods to determine or describe regulatory networks; pathway analysis; …
GBIO0009 - LECTURE 1
Systems Biology This category includes whole cell approaches to molecular biology. Any combination of experimentally collected whole cell systems, pathways or signaling cascades on RNA, proteins, genomes or metabolites that advances the understanding of molecular biology or molecular medicine will be considered. Interactions and binding within or between any of the categories will be considered including protein interaction networks, regulatory networks, metabolic and signaling pathways. Detailed analysis of the biological properties of the systems are of particular interest.
GBIO0009 - LECTURE 1
Becoming a bioinformatician
10 useful / necessary skills
Strong background in some aspect of molecular biology!!!
Ability to communicate biological questions comprehensibly to computer scientists
Thorough comprehension of the problem in the bioinformatics field
Statistics (association studies, clustering, sampling)
Ability to filter, parse, and munge data and determine the relationships between the data sets
GBIO0009 - LECTURE 1
10 useful / necessary skills
Mathematics (e.g. algorithm development)
Engineering (e.g. robotics)
Good knowledge of a few molecular biology software packages (molecular modeling / sequence analysis)
Command line computing environment (Linux/Unix knowledge)
Data administration (esp. relational database concept) and Computer Programming Skills/Experience (C/C++, Sybase, Java, Oracle) and Scripting Language Knowledge (Perl and perhaps Phython)
GBIO0009 - LECTURE 1
“Dammit Jim, I’m a doctor, not a bionformatician!”
(http://www.youtube.com/watch?v=MULMbqQ9LJ8)
The following slides summarize key ideas from the white paper by Christophe
Lambert – 2011, CEO and President of Golden Helix; course website:
Lambert2011
GBIO0009 - LECTURE 1
Bioinformatics-driven genetic research
“your job and expertise is to do x, but to achieve your goals you also have to
do y and z, which you either don’t want to do or don’t have the skills to do”
It takes more than just brains to make productive advances in
bioinformatics:
GBIO0009 - LECTURE 1
Skillset
The free software tools used today require highly skilled bioinformatics
professionals, which are often in short reply …
One must have competences in several disciplines: computer science,
statistics and genetics. Yet, good software overcomes limits in certain
skillsets
In practice though, one virtually has to be a computer programmer in order
to perform genetics research … Why ?
GBIO0009 - LECTURE 1
Toolset
To be able to deal with unavailable informatics expertise that can take advantage of for instance the new sequencing capabilities, some minimal requirements for the toolset are needed: - Robust (not only for local applications) - Well-documented (for application by non-informaticians) - Well-supported (to address questions from users with different
backgrounds) - Maintained and updated (state-of-the art optimized software; is the
case for about 9% of academic programs developed for genetic analysis)
- Multi-platform / Cloud-friendly - Data format transformations (omics integration!)
GBIO0009 - LECTURE 1
Toolset (continued)
Foundational to the problem is the discrepancy between: - The fact that academia is the birthplace of most new statistical and
computational methods in genetic research - When funding ends, software support often ends - Innovation is abandoned (i.e., a new promising tool is not being used)
when one cannot be sure about the accuracy of results (due to lack of documentation about what certain program options imply)
Nevertheless, there are positive examples:
The BIOCONDUCTOR project
GBIO0009 - LECTURE 1
Mindset
Each person has a mindset = the amalgamation of values and mental models of reality with which one engages in a goal-directed behavior. - Values determine goals - Mental models determine means of moving towards these goals
How to measure the distance between current state and goal? - Any such metrics will drive behavior
In academia, the prime metric is reputation: - Recruiting - Promotions - Grant funding
Alternative to fuzzy metric of “reputation”: publications
Publish or perish
GBIO0009 - LECTURE 1
Impact of mindset on toolset
Focus on papers and hence mathematical and algorithm innovation
Ample resources (time, human resources, financial resources) for software quality
Inadequate translation from development to application
GBIO0009 - LECTURE 1
Impact of mindset on toolset
Discussion statement:
“While I do not believe it is intentional, a pattern operates where the scientific
promise of the freely downloadable software is touted in the paper, and then
the “consumer,” due to time pressures and challenges of learning the
software, programming/statistical skillset deficits, and/or fear of improperly
driving the software and making embarrassing mistakes in publications, must
call in the software author and/or his students as consultants on projects to
get the research done.”
GBIO0009 - LECTURE 1
Consequences of the mindset-toolset impact
Bioinformaticians have become the constrained bottleneck resource - Genomic projects too heavily rely on (in-house) bioinformaticians - Because of scarcity, labs may need to wait for months and months to
get the data analyzed, hereby delaying or blocking next steps of the research pro
Outsourcing may be an option (when financial resources are available), but will impact the learning loop: the OODA loop model for goal-directed activities
GBIO0009 - LECTURE 1
Consequences of the mindset-toolset impact
Discussion statement:
“If inadequate or missing documentation, lack of support on appropriate
platforms, buggy or unstable code, and generally low usability are the norm
for academic software, how can one reproduce research when the tools are
unusable? Worse, as quantified earlier, a large fraction of software tools that
are developed and used by a researcher to produce a publication cease to
exist within a few years. Reconstruction of the steps of another researcher
ranges from tedious to impossible.”
In class reading: p5-8 (group discussion)
GBIO0009 - LECTURE 1
Take-home message for the practical side of this course
Checking the quality of analysis results:
“…This could be made possible with software tools that automatically log all
analysis steps and parameters and saves the intermediate steps of analysis,
including graphical output, for inspection at a later date. Further, the ability to
annotate one’s work as one might do in a laboratory notebook would allow an
analyst (expert or novice) to share his or her work with a colleague for
collaboration, review, and audit.”
GBIO0009 - LECTURE 1
3 Gen-omics
Corner stones of bioinformatics
Genetics / Genomics
Genetics is the study of how traits
such as hair color, eye color, and
risk for disease are passed
(“inherited”) from parents to their
children. Genetics influence how
these inherited traits can be
different from person to person.
Your genetic information is called
your genetic code or genome.
Your genome is made up of a
chemical called deoxyribonucleic
acid (DNA) and is stored in almost
every cell in your body.
GBIO0009 - LECTURE 1
Information is everywhere: the programming of life
http://www.youtube.com/watch?v=00vBqYDBW5s
“Information:
that which can be communicated through symbolic language”
GBIO0009 - LECTURE 1
Molecular biology
(adapted from: Davies et al 2009, Integrative genomics and functional explanation)
GBIO0009 - LECTURE 1
Computational biology (“early bioinformatics”)
Biology = noun
Computational = adjective
When I use my method (or those of others) to answer a biological question, I
am doing science. I am learning new biology. The criteria for success has
little to do with the computational tools that I use, and is all about whether
the new biology is true and has been validated appropriately and to the
standards of evidence expected among the biological community. The papers
that result report new biological knowledge and are science papers. This is
computational biology. (https://rbaltman.wordpress.com)
Computational biology = the study of biology using computational techniques. The goal is to learn new biology, knowledge about living sytems. It is about science.
GBIO0009 - LECTURE 1
Bioinformatics
Bio(logy) + Informatics
2 nouns
When I use my method (or those of others) to answer a biological question, I
am doing science. I am learning new biology. The criteria for success has
little to do with the computational tools that I use, and is all about whether
the new biology is true and has been validated appropriately and to the
standards of evidence expected among the biological community. The papers
that result report new biological knowledge and are science papers. This is
computational biology. (https://rbaltman.wordpress.com)
Bioinformatics = the creation of tools (algorithms, databases) that solve problems. The goal is to build useful tools that work on biological data. It is about engineering.
GBIO0009 - LECTURE 1
Genetic epidemiology
Genetic epidemiology is a particular sub-discipline of epidemiology which
considers genetic influences on human traits.
As with any other epidemiologic studies, genetic epidemiology studies aim
- to assess the public health importance of diseases,
- to identify the populations at risk,
- to identify the causes of the disease,
- and to evaluate potential treatment or prevention strategies based on
those findings
Strategies of analysis include population studies and family studies.
- Huge challenge is to combine big data repositories from population or
family-studies (e.g., genomics, transcriptomics, metagenomics,
metabolomics, epigenomics, ….) with clinical and demographic data:
BIOINFORMATICS
GBIO0009 - LECTURE 1
Principal internet resources for genome browsers and databases
Resource Web address Description Sponsoring
organizations
Open Helix http://www.openhelix.com/tutorials.shtml
On-line tutorial material for all
of the genome databases. OpenHelix, LLC
UCSC
Genome
Browser
http://genome.ucsc.edu
Comprehensive, multi-species
genome database providing
genome browsing and batch
querying.
Genome Bioinformatics
Group, University of
California, Santa Cruz
Ensembl
Browser http://www.ensembl.org
Comprehensive, multi-species
genome database providing
genome browsing and batch
querying.
European
Bioinformatics Institute
(EBI) and the Sanger
Center
NCBI
MapViewer http://www.ncbi.nlm.nih.gov/mapview
Multi-species genome browser
focusing especially on genome
mapping applications.
National Center for
Biotechnology
Information (NCBI)
GBIO0009 - LECTURE 1
Biomart http://www.biomart.org/
Genome-database, batch-querying
interface used by Ensembl and
several single-genome databases.
Ontario Institute for Cancer
Research and European
Bioinformatics Institute
Galaxy http://main.g2.bx.psu.edu
Integrated toolset for analyzing
genome batch-querying data.
Center for Comparative Genomics
and Bioinformatics. Penn State
University
Taverna http://taverna.sourceforge.net
Toolset for creating pipelines of
bioinformatics analyses implemented
via the Web services protocol.
Open Middleware Infrastructure
Institute, University of
Southampton (OMII-UK)
GMOD http://www.gmod.org
Repository of software tools for
developing generic genome
databases.
A consortium of organizations
operating as the Generic Model
Organism Database project
(Schattner et al. 2009)
GBIO0009 - LECTURE 1
R (TA- sessions)
R is a freely available language and environment for statistical computing
and graphics which provides a wide variety of statistical and graphical
techniques: linear and nonlinear modelling, statistical tests, time series
analysis, classification, clustering, etc.
Consult the R project homepage for further information.
The “R-community” is very responsive in addressing practical questions
with the software (but consult the FAQ pages first!)
CRAN is a network of ftp and web servers around the world that store
identical, up-to-date, versions of code and documentation for R.